Transverse Disciplines in Metrology: Proceedings of the 13th International Metrology Congress, 2007 - Lille, France

By French College of Metrology

Based on The International Metrology Congress meeting, this reference examines the evolution of metrology, and its applications in industry, environment and safety, health and medicine, economy and quality, and new information and communication technologies; details the improvement of measurement procedures to guarantee the quality of products and processes; and discusses the development of metrology linked to innovating technologies. The themes of the Congress (quality and reliability of measurement, measurement uncertainties, calibration, verification, accreditation, sensory metrology, regulations and legal metrology) are developed either in a general way or applied to a specific economic sector or to a specific scientific field.

• Language: English
• Publisher: Wiley
• Release date: May 10, 2013
• ISBN: 9781118623411

Book preview

Certification of a reference material for herbicides in water by way of inter laboratory comparison

B. Lalèrea, V. Le Diourona, M. Désenfanta,K. El Mrabeta,b V. Pichonb, J. Vialb, G. Hervoüeta

a Laboratoire National de Métrologie et d’Essais (LNE), 1 rue Gaston Boissier, 75724 Paris cedex 15, France

b Laboratoire Environnement et Chimie Analytique (LECA) UMR-CNRS 7121, Ecole Supérieure de Physique et Chimie Industrielles, 10 rue Vauquelin, 75231 Paris Cedex 05, France

---

ABSTRACT: The certified reference materials (CRMs) are among the most appropriate tools for the traceability and validation of analytical methods. Although their number seems to be high 20,000, they only cover a small part of the analytical needs for environmental monitoring. Until now, none were available for analyzing pesticides in water.

After an intralaboratory feasibility study and an evaluation of the behavior of this material during interlaboratory testing, the LNE has produced and certified a CRM for the triazine and phenylurea analysis in water samples.

---

Introduction

In accordance with the European outline directive on water (DCE), the department of ecology published memorandum service instruction DCE 2006/16 relating to the constitution and implementation of a monitoring program for the different water categories. Because of this, quality of rivers and ground water is regularly monitored. Pesticide content, a micropollutant considered as a priority by the European Union in particular is verified.

In order to achieve this monitoring, several measures are taken daily by different laboratories. It is important to be able to compare results. This objective can only be reached by ensuring traceability of these analyses with the help of different tools such as certified reference material (CRM), the link to national standards through uninterrupted chains of comparison and a participation in interlaboratory testing. CRMs have the advantage of also being able to evaluate accuracy during validation of analytical methods implemented. They can be of two types: those destined to calibrate measurement systems and matrices making it possible to consider all steps from the preparation of the sample.

Until now, there were none for the analysis of pesticides in water. That is why LNE in cooperation with LECA decided to develop one. This project started four years ago with a feasibility study of such a material. The behavior of products developed was then evaluated during an interlaboratory comparison.

A group of CRMs was developed in March 2006 and certified in June 2006 for the analysis of triazines and phenylureas in drinking water.

Previous studies: feasibility and behavior of RM

The following components which belong to two pesticide families were selected:

– triazines: deisopropylatrazine (DIA), deethylatrazine (DEA), simazine, atrazine, terbutylazine and terbutryne.

– phenylureas: chlortoluron, diuron, isoproturon and linuron.

Completed in 2002, this choice, except for terbutylazine and DIA, is based on the frequency of detection of these triazines and phenylureas in water [1]. In addition, they are indexed in the list of 50 pesticides sold in larger quantities than 500 tons per year in Europe. Terbutylazine, following the ban on the use of atrazine, is part of the mix used for its replacement in corn cultures and thus was added. Despite its usage ban, atrazine can still be detected in water; because of this, two of its metabolites, DEA and DIA were included in this study.

Feasibility study

In order to achieve the most complete kit possible, two types of reference materials (RM) were retained:

– sealed vials: targeted pesticides are stored:

- in a solution of acetonitrile,

- dried after evaporation of the solvent used for their preparation.

– solid phase extraction (SPE) cartridges called cartridges in this document: a water sample spiked with each compound is percolated on the support in order to retain them with other molecules in the sample. Two polymeric supports were evaluated: a divinylbenzene copolymer functionalized with N-vinylpyrrolidone (Oasis HLB, Waters) groups and a polystyrene divinylbenzene copolymer (ENVI Chrom P, Supelco). Analysis laboratories will then have to carry out the elution of compounds.

These materials were prepared with two levels of concentration, targeted in relation to regulation (0.15 μg/l for drinking water and 0.50 μg/l in surface water) T=-18°C for a year. Every month, the evolution of compound concentration was studied by the analysis of vial and cartridge contents.

The detailed presentation of results for each compound, each type of material and each storage condition was the subject of publication [2]. Between the beginning and ending of the study, the evolution was quantified in percentage of the quantity found in relation to the one initially introduced (rate of recovery). To summarize, compounds are classified into three families of behaviors:

– non-usable family (Figure 1): either the compounds were totally degraded or their recovery level after nine months of study are lower than 10%;

– family with trend (Figure 2): compound concentration evolves over time;

– family without trend (Figure 3): pesticide concentration does not evolve significantly over time.

Figure 1. Non-usable family example: Linuron stored in vial after dry evaporation and conservation at ambient temperature

Figure 2. Family with trend example DIA stored in vial after dry evaporation and conservation at ambient temperature

Figure 3. Family without trend example Atrazine stored in vial after dry evaporation and conservation at T=-18°C

The results have shown that compounds are stable when then are in a solution of acetonitrile when temperature becomes ambient, whereas it is necessary to store them at -20°C when they are dried in a vial. When they are fixed on cartridges, storage temperature must be lower than 0.5°C. Observed behaviors are identical regardless of the level of concentration.

The retained RM was then a kit with:

– vials containing pesticides in acetonitrile;

– cartridges (Oasis HLB, Waters) treated by percolation of drinking water containing selected pesticides.

As a precaution, they will be stored at -20°C.

RM behavior during an interlaboratory comparison

In order to test the conditions of RM usage (sending conditions, reception and analysis with different methods), an interlaboratory test involving about 15 participants was organized [3].

The variation factors of both components of the material tested vary between 14 and 30% depending on compounds, which is very satisfactory for a circuit involving laboratories using different methods compared to other campaigns [4,5].

In view of the results (feasibility and interlaboratory test), a certification campaign was then completed.

Material production and certification

This reference material, in the form of a kit is made up of:

– two vials containing approximately 1.2 ml of a herbicides solution (atrazine, deethylatrazine, deisopropylatrazine, simazine, terbutryne, terbutylazine, chlortoluron, linuron, diuron, and isoproturon) in acetonitrile (concentration for each pesticide ≈ 0.1 mg/l);

– three cartridges on which 250 ml of a drinking water spiked with a solution of herbicides in acetonitrile was percolated, in order to reach a concentration of approximately 0.50 μg/l of each compound in water.

This material is destined to calibrate the measurement devices and/or to validate the analytical methods for the determination of herbicides in water.

Preparation of materials

The preparation of materials required the creation of multi compound solutions by consecutive dilutions of a mother solution obtained with pure compounds.

After preparation, the solution is transferred into vials which are immediately sealed under nitrogen.

For the cartridges, 180 l of tap water were taken the same day. Before transferring to a cartridge according to the protocol described below, 5 l of water are spiked with 5 ml of a pesticide solution.

Cartridges are first conditioned with 3 ml of acetonitrile, 3 ml of methanol, 3 ml of water then 250 ml of spiked water are percolated. Water rinsing and nitrogen flow drying steps are then realized before storing them in specific conditions (protected from light and at a temperature lower than –20°C). All these steps were conducted with the ASPEC XL IV robot (GILSON).

There was a prior verification that vial seals would not alter the solution inside and that the reliability of the robot was sufficient to guarantee good cartridge homogenity.

Vial and cartridge preparations required three days and three weeks respectively.

400 sealed vials and 600 cartridges were produced.

Homogenity

The homogenity was verified on two series of ten vials and ten cartridges (Table 1).

Table 1. Vial and cartridge homogenity

Stability

During the life span of the material, analyses are done each month. Until now and in accordance with the feasibility study, no significant evolution of concentrations was detected as is shown for example in Figures 4 and 5.

Figure 4. Stability of chlortoluron in solution in the vials

Figure 5. DEA stability in cartridge

Inter laboratory testing

16 laboratories with different techniques participated in this test (Table 2).

Samples (3 vials and 3 cartridges) were sent on April 4th 2006 and all results were received on May 20th 2006.

Table 2. Laboratories participing in the certification testing and implemented techniques

Results

Raw results for cartridges and vials are grouped in Tables 3 and 4 respectively.

Table 3. Laboratory results for cartridges expressed in μg/l of water

Table 4. Laboratory results for vials expressed in mg/l

Statistical data

Statistical data analysis

The statistical data analysis was performed based on operation standards for interlaboratory tests NF ISO 5725-2 [6] and NF ISO 5725-5 [7]. The first phase was to detect atypical values from statistical homogenity tests (Grubbs and Cochran). The exclusion of some data was based on statistical and technical conclusions. The second phase, from results obtained, was to quantify the parameters summarizing the RM, average and standard deviation of reproducibility as closely as possible. The results of this process are summarized in Tables 5 and 6.

Table 5. Statistical data for vials

Table 6. Statistical data for cartridges

Certified values

The value assigned to CRM is the average result from laboratories and the standard uncertainty is given by the standard deviation of reproducibility. This uncertainty corresponds to the uncertainty on the result of an analysis of this CRM carried out by a laboratory working under the same conditions as the circuit laboratories.

Tables 7 and 8 group CRM certified values.

Table 7. Certified values for vials

Concentrations in cartridges are expressed in μg of compound by liter of water. In each cartridge, 0.25 l of water was percolated.

Table 8. Certified values for cartridges

Conclusion

Validation for this experimental approach makes it possible to propose a certified reference material which is now available for the analysis of pesticides in water. It is presented in the form of a kit in order to respond to analysis laboratory requirements:

– sealed vials containing compounds in solution in acetonitrile, which can be used for the verification of calibration or for spiking water matrices;

– cartridges on which water containing compounds was percolated representing a real sample.

From the development to certification of this reference material, the study took five years of study. In addition, this CRM is in compliance with the ISO 34 guide specifying the conditions of production, sample conservation and stability.

References

[1] Etudes et Travaux IFEN, Les pesticides dans les eaux, September 2000.

[2] J. Deplagne, J. Vial, V. Pichon*, Béatrice Lalere, G. Hervouet and M.-C. Hennion, Feasibility study of a reference material for water chemistry: Long term stability of triazine and phenylurea residues stored in vials or on polymeric sorbents, Journal of Chromatography A, 1123 (2006) 31-37.[3] K. El Mrabet, M. Poitevin, J. Vial*, V. Pichon, S. Amarouche, G. Hervouet, B. Lalere, An Interlaboratory Study to evaluate potential Matrix Reference Materials for pesticides in Water, Journal of Chromatography A, 1134 (2006) 151-161.

[4] S.A. Senseman, T.C. Mueller, M.B. Riley, R.D. Wauchope, C. Clegg, R.W. Young, L.M. Southwick, H.A. Moye, J.A. Dumas, W. Mersie, J.D. Mattice, R.B. Leyy, Interlaboratory comparison of extraction efficiency of pesticides from surface laboratory water using solid-phase extraction disks, J. Agric. Food Chem., 51 (2003) 3748-3752.

[5] M.B. Riley, J.A. Dumas, E.E. Gbur, J.H. Massey, J.D. Mattice, W. Mersie, T.C. Mueller, T. Potter, S.A. Senseman, E. Watson, Pesticide extraction efficiency of two solid phase disk types after shipping, J. Agric. Food Chem., 53 (2005) 5079-5083.

[6] NF ISO 5725-2 Exactitude (justesse et fidélité) des résultats et méthode de mesure partie 2: Méthode de base pour la détermination de la répétabilité et de la reproductibilité d'une méthode de mesure normalisée

[7] NF ISO 5725-5 Exactitude (justesse et fidélité) des résultats et méthode de mesure partie 5: Méthodes alternatives pour la détermination de la fidélité d'une méthode de mesure normalisée (analyses robustes).

---

Determination of aflatoxin M1 using liquid chromatographic method: comparison between uncertainty obtained from in house validation data and from proficiency test data

C. Focardi, M. Nocentini, S. Ninci, P. Perrella, G.Biancalani

Istituto Zooprofilattico Sperimentale delle Regioni Lazio e Toscana - Sezione di Firenze, Via di Castelpulci 43, 50010 San Martino alla Palma, Florence, Italy. Tel [phone]- Fax [phone] – e-mail: [e-mail]

---

ABSTRACT: An HPLC method with fluorescence detection has been validated for the determination of aflatoxin M1 in milk samples. Certified Reference Materials and Spiked samples have been used for in house validation. These validation data have been applied for the uncertainty estimation with the bottom-up approach. Results obtained with this method have been compared with the expanded uncertainty determined from proficiency testing FAPAS.

---

Introduction

Aflatoxins are a group of hepatocarcinogen molecules produced by Aspergillus flavus and Aspergillus parasiticus. When aflatoxin B1 (AFB1), present in contaminated feed, is ingested by dairy cattle, it is excreted into milk as aflatoxin M1 (AFM1). Both AFB1 and AFM1 can cause DNA damage, gene mutation, chromosomal anomalies and as a consequence the International Agency for Research on Cancer (IARC) has classified AFB1 and, recently also AFM1, as class 1 (carcinogenic to humans) [1]. Strict regulatory limits for AFM1 are currently in force in the European Community; the Regulation (EC) 1881/2006 [2] set a maximum residue level (MRL) of AFM1 in milk, intended for adults, at 0.050 μg/kg, and at 0.025μg/kg for milk, intended for infants or for baby-food production. The European Commission with the Regulation (EC) 401/2006 [3] fixed the performance criteria for the analytical methods.

The adopted method is a liquid chromatographic one with fluorimetric detection. The sample is purified by using immunoaffinity column. The method is in house fully validated using Certified Reference Materials and spiked samples.

The EN ISO/IEC 17025 [4], as well as the Regulation (EC) 401/2006, requires all the measurements to be accompanied by estimation of expanded uncertainty.

Various methods are available to evaluate measurement uncertainty. One approach, applied by Eurachem [5], is the so called bottom-up and consists in separately identifying and quantifying error components that are consider important from in house validation data. Another approach is to use results from proficiency testing, by comparing reproducility variance with the repeteability variance of the laboratory [5].

In our case we use the data obtained with participation of the laboratory at 11 FAPAS during the period 2002-2006.

Experimental

Materials

Aflatoxin M1 Standard Reference Material at a concentration of 500 μg/l in methanol was purchased from Riedel de Haën. The chemical structure of AFM1 is presented in Figure 1.

Figure 1. Chemical structure of Aflatoxin M1

Deionised water obtained by a Milli-Q water (Millipore) and acetonitrile HPLC grade where used throughout. AflaTest® Immunoaffinity columns containing antibodies against AFM1 were purchased from VICAM. AFM1 working standard solutions at five different concentration levels (0.08, 0.2, 0.4, 0.6 and 0.8 ng/ml) were prepared by dilution of the stock solution (5 ng/ml). All the solutions are dissolved in acetonitrile water in the ratio 10/90 (v/v) (Figure 2).

Sample preparation

The reference materials used for HPLC method validation and uncertainty estimation are reported in Table 1.

Table 1. Reference materials used for method validation

These milk powder samples were prepared by dilution with water in the ratio 1:10 (w/w), in the way that the final concentration of AFM1 in the samples lays down in the HPLC calibration curve. Besides spiked samples at the nominal concentration of 0.05, 0.04 and 0.03 μg/Kg were prepared adding, under a gently magnetic stirring, respectively 500 μl, 400 μl and 300 μl of AFM1 stock solution (5ng/ml) to 50 g of homogenized milk, previously tested to demonstrate the absence of AFM1 residues, under a gently magnetic stirring.

Reconstituted Certified Materials (FAPAS and CRM), spiked and blank milk samples were extracted and purified with the procedure used by Tuinistra [6] and reported in the flow diagram (Figure 2).

Figure 2. Flow diagram of sample analysis

To ensure the homogenity of samples, they were gently magnetically stirred at room temperature for about 15 minutes. After homogenization, 50 g of samples were centrifuged for 25 minutes at 3500 rpm, to eliminate fat components. Skimmed milk was passed through the immunoaffinity column. The column was washed with 10 ml of water and AFM1 eluted with 5 ml of acetonitrile.

For chromatographic analysis the sample was evaporated under a stream of nitrogen to a volume of approximately 0.3 ml and reconstituted to 5 ml with water.

HPLC analysis

Chromatography was performed with a Perkin Elmer (USA) Series 200 system consisting of a quaternary gradient pump, an autosampler, a fluorescence detector and a degassing system using helium. Chromatographic separation was achieved using a Spherisorb ODS2 (250 × 4.6 mm, 5 μm) reversed phase column, with a guard column of the same type. The mobile phase was constituted by water and acetonitrile in the ratio 75/25 (v/v). An isocratic elution was performed at a flow rate of 1 ml/min for a total run time of 20 minutes. The injection volume was 500 µl. The detector was set at a excitation wavelength of 363 nm with a emission wavelength of 433 nm.

HPLC validation criteria and uncertainty estimation

The method for the determination of Aflatoxin M1 in milk was in house validated by a set of parameters which are in compliance with the Eurachem guide [7] and the Commission Regulation 401/2006 [3].

For evaluating the overall uncertainty, the bottom up approach has been adopted, following the statements of Eurachem guide [5].

Internal quality control

The method used for the determination of aflatoxin M1 in milk fulfils the requirements of EN ISO/IEC 17025 [4] and is accredited in the laboratory.

Internal quality control is achieved following the IUPAC harmonized guidelines [8]. A HPLC sequence consists of the analysis of a standard, a blank sample, a Certified Reference Material to check the recovery and finally samples in duplicate.

The Shewart control chart has been applied for the statistical control of the measurement process by using the CRM and an example is reported in Figure 3. The Reference value μ is the concentration level of CRM and σ is the uncertainty associated with the material. The warning limits are considered equal to μ plus and minus 2σ, the action limits are μ plus and minus 3σ below the center line.

Figure 3. Shewart control chart related to FAPAS 0445

Proficiency testing

The performance of the laboratory is periodically checked with the partecipation at Proficiency testing organized by FAPAS®. The results were evaluated in the form of a z-score which is the estimated laboratory bias divided by the target value for the standard deviation [8].

The laboratory has participated in eleven proficiency tests over the years. In Figure 4 we show the value of z-score obtained according to time (omissis); all values are included in the range of ± 1.

Figure 4. z-score profile obtained during the period 2002-2006

Results and discussion

In-house validation of the method

The validation data are obtained by using Certified Reference Materials and spiked samples.

Specificity. The chromatograms of a blank milk sample, CRM and standard solution are previously reported [9]. The blank milk sample is free of interferences at the elution time corresponding to the AFM1 peak, demonstrating a good specificity of the method proposed.

Linearity. The calibration curve of AFM1 in standard solutions (Figure 5) is linear (r = 0.9999) over the concentration range between 0.08 and 0.8 ng/ml. These data are the results of six different series obtained by different operators.

Figure 5. Calibration curve for aflatoxin M1

Evaluation of systematic error. The Bartlett test applied both on Certified Reference Material and spiked samples demonstrated the homogenity between the variance and subsequently the absence of systematic errors.

Accuracy and precision. The precision and accuracy determined are listed in Table 2.

According to the EC Regulation 401/2006, the accuracy for all of the samples fell in the range between –20 % to 10%. The precision of the method is expressed as RSD, Relative Standard Deviation, values for all concentrations. In Table 2 are also reported the values of the RSDmax which is equal to two thirds of the Horwitz equation:

where C is the mass fraction expressed as a power of 10. According to the EC Regulation 401/2006 experimental RSD value is lower than RSDmax.

Table 2. Data of accuracy and precision obtained for CRM, FAPAS and Spiked samples, where n is number of replicates and Mean is the average concentration obtained

Limit of detection and limit of quantification. The limit of detection has been calculated from the calibration curve. Taking into account the standard deviation (Sy/x) of the regression analysis, the limit of detection is found to be CLOD= 0.001 μg/kg. The limit of quantification (LOQ) has been set as 10-fold the LOD giving the value of CLOQ= 0.01 μg/kg.

Uncertainty estimation

The equation of the measurand is as follows:

(1)

where M (g) is mass weight of the milk sample, Vext is the final volume of the extraction, Cx (ng/ml) is the concentration obtained by calibration curve and R is the recovery rate obtained on suitable reference sample (CRM or FAPAS sample). Every factor of the equation is shown in the cause and effect diagram (Figure 6).

Figure 6. Cause and effect diagram

The detail of uncertainty estimation, based on bottom-up approach taking into account the in-house validation data are previously reported [9]. For the standard uncertainty associated to the dilution volume and weight, a triangular distribution has been chosen.

In Figure 7 is reported the Error Budget diagram; the contribution to uncertainty of mass weight (M) and dilution (Vext) are negligible compared to the others, for all samples analysed.

Figure 7. Error budget diagram

In Figure 8 are presented the pie charts for the reference materials and for the spiked samples.

If we compare the different components to uncertainty of certified reference materials to those obtain for the spiked samples, it is clear that in the second case increases the contribution of the recovery, due to the homogenity of the sample. At the same time, the contribution due to the concentration Cx decreases. This second effect should be correlated with the value of the chromatographic peak area which is lower in the case of Certified Reference Material, due to the dilution of the sample.

Figure 8. Contributions to combined standard uncertainty. Charts show the relative sizes of uncertainty associated with precision, bias, calibration and other effects

According to the bottom up approach the global uncertainty, is given by the addition of all the uncertainty associated with the component that influences the measurand. Taking into account equation (1) the overall uncertainty, in terms of relative uncertainty, is determined by the formula:

The standard uncertainty is then obtained by the equation

(2)

In Table 3 we show, for the Certified Reference Materials and for the spiked sample, the relative uncertainty ů(C).

Table 3. Relative and standard uncertainty calculated for reference materials and spiked samples

The relative expanded uncertainty is obtained multiplying the mean value of the relative uncertainty, by a coverage factor, k=2

As an alternative method to measure uncertainty for chemical measurements, the Analytical Method Committee of the Royal Society of Chemistry [10] proposed an approach based on precision data assessed in an interlaboratory study.

The uncertainty associated with a mesurand result y is given by the following formula:

where s²R is the reproducibility variance between laboratories and uref is uncertainty associated with the accepted reference value. A further example of this expression is possible when u²ref is negligible in comparison to the reproducibility variance.

Some applications of interlaboratory data in the estimation of the uncertainty of chromatographic methods have been described [11,12]. Considering the results obtained by the laboratory, applying the proficiency test FAPAS, we can compare laboratory repeatability with the reproducibility of interlaboratory study (Table 4).

Table 4. Results of FAPAS proficiency testing: C is Assigned value and σ is the robust standard deviation, σrel is the relative robust standard deviation

The mean value obtained for the intralaboratory relative repeatability is equal to 0.09, an order of magnitude lower than the interlaboratory reproducibility. This indicates that the method precision of the laboratory is comparable to that of the laboratories which took part in the collaborative trial. It is therefore acceptable to use the reproducibility standard deviation from the collaborative trial in the uncertainty budget of the method. The mean value resulting from the interlaboratory test of relative robust standard deviation is equal to 0.28.

We can calculate the expanded uncertainty at the MRL level by using the relative uncertainty obtained with the two different approaches applying equation (2).

We can conclude that the expanded uncertainty obtained with the two different methods are of the same order of magnitude.

Conclusions

This chromatographic method for the determination of aflatoxin M1 in milk samples meets compliance with the Regulation (EC) 401/2006. Internal quality controls are achieved by applying the Shewart control chart to the data obtained each round with the Certified Reference Materials. Uncertainty has been evaluated following two different methods; the bottom up approach and the interlaboratory data, in the specific case, FAPAS round materials over the period 2002-2006. Expanded uncertainty obtained with the two different approaches calculated at the Maximum Residue Level are comparable.

References

[1]IARC International Agency for Research on Cancer Monograph on evaluation of carcinogenic risks to humans. IARC Lyon France 2002 82, 171

[2]Official Journal of European Commission L 364/5 2006 Commission Regulation No 1881/2006 of 19 December 2006. Brussels Belgium

[3]Official Journal of European Commission L 70/12 2006 Commission Regulation No 401/2006 of 23 February 2006. Brussels Belgium

[4]EN ISO/IEC 17025: 2005. General Requirements for the competence of calibration and testing laboratories. 2005 ISO Geneva

[5] Quantifying uncertainty in analytical measurement Eurachem Guide 2000, second edition LGC.

[6]Tuinistra L. et al., J. AOAC Int, 1993, 84, 2. 1248-1254

[7]The Fitness for purpose of analytical methods A laboratory guide to method validation and related topics Eurachem Guide 1998, edition 1.0 LGC.

[8] Thomson M., Pure Applied Chem, 1995, 4, 649-666

[9]M. Nocentini, C. Focardi, M. Vonci, F. Palmerini. Proceedings at 11thInternational Metrology Congress, 20-23 October 2003 Toulon- France

[10] AMC (Analytical Method Committee) Analyst, 2005, 130, 721

[11] Dehouck P., et al., J. of Chrom A, 2003, 1010, 63-74

[12] Dehouck P., et al., Anal Chim Acta, 2003, 481, 261-272

---

Purity analysis in gas metrology

F. Dias and G. Baptista

Laboratório de Gases de Referência, Instituto Português da Qualidade R. António Gião, 2, 2829-513 Caparica, Portugal

---

ABSTRACT: The Reference Gas Laboratory (LGR) of the Portuguese Institute for Quality (IPQ) is the Primary Laboratory in Portugal in the field of gas metrology. Its main mission is to assure and guarantee the accuracy and traceability [1] of the gas measurements to national and international standards. LGR is also a consumer of pure gases to be used as raw material in the gravimetric preparation of primary gas mixtures. The laboratory needs to measure and control the purity level of the gases supplied by the industry in order to determine the gas composition with higher accuracy. Several methods are used for purity analysis, namely, Gas Chromatography (GC), Fourier Transform Infrared Spectrometry (FTIR) and Mass Spectrometry (MS). These analytical methods will be briefly described and documented with some examples of gas purity analysis made at LGR: Propane (C3H8), Nitrogen Monoxide (NO) and automotive mixtures (CO+CO2+C3H8 in Nitrogen).

---

Introduction

LGR is a consumer of pure gases to be used as raw material in the preparation of primary gas mixtures. The need for the quantification of impurities is of high interest in the calculation of the measurement uncertainty and in the estimation of the result accuracy.

The choice of the more adequate analytical technique is made according to different properties such as, method specificity, cross-interference and matrix effects and also the method detection limits.

At LGR, several methods are being implemented for purity analysis, namely, Gas Chromatography (GC), Fourier Transform Infrared Spectrometry (FTIR), and Mass Spectrometry (MS). In GC technology, the purity of parent gases is obtained by using the two following techniques: Flame Ionization Detection (FID) and Thermal Conductivity Detection (TCD). The FTIR technique is used to measure gas species, which may be difficult for GC analysis. GC-MS technology is used to detect multi-species by mass signatures and identify the unknown impurities.

In order to obtain a better knowledge of the purity of gases involved in gravimetric preparations and/or in dynamic mixtures, our laboratory is now developing a Purity Analysis service. Although not fully implemented these techniques already provide important information concerning raw material control, fundamental to primary gas standards preparation. The quantification of impurities is of high interest in the calculation and estimation of the results. Indeed, one of the major uncertainty contributions is the purity component of the gas.

The qualitative analysis of impurities in pure gases is also important due to the possibility of having interferents and because of the need to ensure that crossinterferent impurities are small enough not to contribute significantly to the measured results.

The gas mixture preparation method according to ISO 6142 accounts for each component (including impurities) as a fraction amount of the total mixture. It then evaluates the uncertainty of each fraction amount. This approach is totally correct, but it might lead to neglecting certain correlations in the uncertainties when the same source of gas is used more than once in the preparation of a mixture.

Gas Chromatography (GC)

Gas Chromatography (GC) [2] is a common analytical method for gas purity analysis. GC is able to analyze almost all gas components, but requires a variety of columns, specific for certain chemical species, and detectors. This technique has low gas consumption and it is possible to connect it to auto-sampler equipment enabling automatic data acquisition. One limitation is the need for reference standards to validate peak retention time. It is a good tool for quantitative analysis.

In the LGR two different detectors are used, namely, Thermal Conductivity Detector (TCD) and Flame Ionization Detector (FID), which will be briefly described below.

Figure 1. Gas Chromatograph Agilent 5890

Thermal Conductivity Detector (TCD)

Gas Chromatography by TCD is a non-destructive method in which the detector is sensitive to the thermal conductivity of the carrier gas. Each time a component different from the carrier gas passes through the detector, the TC decreases the given origin to a signal proportional to that component concentration. One important advantage is that it is valid for almost all components.

Flame Ionization Detector (FID)

The flame Ionization Detector (FID) is a destructive method on which the carrier gas is mixed with hydrogen and air, being the final mixture burned. The burning produces ions causing an electric current change. The signal generated is proportional to the ion concentration. It is a very sensitive method allowing the detection limit in GC to be decreased. It applies to almost all organic compounds, but, is not sensitive for common inorganic species (CO, NOx, SO2, H2S, H2O). One way to change this is using a Nickel catalyst, which converts the CO and CO2 to CH4 for further detection.

Pure C3H8 analysis

An example is shown of pure Propane (99.95%) analyzes by GC-TCD with different columns.

Equipment: Agilent 5890

Column: Porapack Q, Molecular sieve 5A

Temperature: 160ºC, 150ºC

Figure 2. C3H8 pure sample analysis with Porapak col

Figure 3. C3H8 pure sample analysis with Mol. Sieve col

C3H8 99.95%

Specification: H2O (5 ppm), O2 (10 ppm), CO2 (5 ppm), N2 (40 ppm), C3H6 (200 ppm), CnHm (200 ppm)

Results: H2O (?), O2 (8 ppm), CO2 (2 ppm), N2 (83 ppm), CnHm (42 ppm CH4 + 21 ppm C2H6)

For the C3H8 99.95% sample, the results show that the mixture is under the specification, except for N2 where it was found at a concentration higher than specified. Nevertheless, it was not possible to separate propene from propane, nor to measure H2O with this technique

Fourier Transform Infrared Spectrometry (FTIR)

FTIR [3] is a spectroscopy method where working principle is based on the infrared (IR) absorption by molecules. Any molecular vibrations, which displace an electric charge, will absorb IR radiation. On FTIR the scanning of IR region allows us to detect several components in a sample. However, it applies only to species absorbing an IR radiation, which means that atomic species (He, Ne, Ar), as well as the homonuclear diatomic species without a permanent dipole (H2, N2, O2) cannot be observed.

This limitation does not imply that the majority of environmental and pollutant gases cannot be observed by this technique. According to NIST, approximately 100 of the hazardous air pollutants listed in the US EPA Clean Air Act can be measured.

Figure 4. Fourier Transform Infrared Spectrometer BOMEM MB100

IR signatures are easily recognized and do not change according to the mixture, showing no matrix effects. The main limitation of this method is the difficulty to eliminate residual H2O (1325-1900 cm-1: 3550-3950 cm-1) and CO2 (2295-2385 cm-1) existing in the absorbance spectrum, mainly due to the evolving atmosphere. The large amount of gases consumed during an analysis is also a disadvantage when the sample quantity is a limitation. For all the reasons explained above, FTIR is mainly used to measure gas species difficult to measure in GC. This technique is a useful tool to measure pure NO and its contaminants.

Pure NO analysis

An example of pure NO 99.90% analysis by FTIR is shown.

Equipment: BOMEM MB100

Scanning Region: Mid Infrared (4000-400 cm-1)

Resolution: 8 cm-1

Gas Cell: Graseby Specac glass cell

Optical Path: variable path length (1-8 m)

Figure 6. Absorbance vs wavenumber (cm-1) FTIR analysis

NO 99.90% specification

Impurities: H2O (20 ppm), NO2 (100 ppm), CO2 (100 ppm),

N2O (200 ppm), N2 (500 ppm),

In the example, the NO 99.90% pure sample, is not in accordance with the specification, namely for the NO2 (590 ppm) and N2O (420 ppm) components which are significantly greater. This difference can be due to the reactivity of NO. However, it is an example of the lack of accuracy in analysis made by some gas suppliers not traceable to the SI.

Gas Chromatography – Mass Spectrometry (GC-MS)

GC-MS [4] is a hyphenated technique, which combines the Chromatographic separation with the spectral information. The GC separates mixtures into their components which will then pass through the MS [5]. Here, each component is fragmented into several ions, by the ion source. The mass filter or quadrupole classifies the different ions into the mass-to-charge ratio (m/Z). The detector will further produce a signal proportional to the ion concentration. GC-MS is very useful for identifying unknown compounds, namely trace contaminants that can be found in pure or balance gases. When there are no clues or reference standards to compare with an unexpected signal appearing in a GC analysis, this analytical technique can give a result, by comparing the mass signature obtained with the mass spectrum library, although the accuracy of quantification is poor.

Figure 7. Gas Chromatograph – Mass Spectrometer Agilent 6890

Automotive exhaust gas analysis

An example of an automotive gas analysis of a Certified Reference Material (CRM) provided by a gas supplier is presented. This CRM contained an unexpected impurity.

Equipment: Agilent 6890

Column: HP PlotQ, HP Molesieve

Temperature: Variable (60–150ºC)

Figure 8. Chromatogram for an automotive CRM gas sample

Figure 9. Mass spectrum comparison: above, unknown component sample; below, ethylene oxide

Automotive exhaust gas CRM specification

Composition: CO (0.5%), CO2 (6%), C3H8 (100 ppm), N2 (matrix)

Comparing the mass spectrum of the unknown component sample with the mass spectrum of Ethylene Oxide (C2H4O), it was observed that the contaminant corresponds to this component. Besides not interfering with components in the mixture, this molecule could introduce a considerable source of error if it was, as previewed, used as standard material for calibration of automotive gas analysis equipment, since this apparatus reads total hydrocarbons.

Conclusions

In gas purity analysis there is no universal equipment. The maximum information on a sample characterization will be given by combining different techniques.

To obtain a quantitative analysis there is the need to purchase all the correspondent reference standards, which is not always easy due to price or availability restrictions.

Gas purity analysis is crucial on primary standards preparation enabling us to decrease measurement uncertainty and therefore reach more accurate results.

References

[1]BIPM et al., International Vocabulary of Basic and General Terms in Metrology, 2ª ed, Geneva, ISO, 1993. 59 p. ISBN 92-67-01075-1.

[2]Agilent 6890 Series Gas Chromatograph Operating Manual, USA, 2000.

[3]The Michelson Series FT-IR Spectrometer – BOMEM User’s Guide, Version 1.0, Canada, 1994.

[4]Agilent GC-MSD Chemstation and Instrument Operation basique H 872090000. USA, 2005.

[5 Agilent Technologies. 5973 Inert Mass Selective Detector Hardware Manual. USA, 2003.

Comparison of two different approaches to evaluate the on line gas chromatography for natural gas

Elcio Cruz de Oliveira

Petrobras Transporte S.A., Av. Presidente Vargas 328, 7o andar – Rio de Janeiro – RJ, ZIP 20021-060, Brazil; Tel + 55 21 3211 9223; Fax + 55 21 3211 9300

Email: [e-mail]

---

ABSTRACT: Recently, several approaches to evaluate the uncertainty in measurement have been developed. Within these, we may highlight the following: the guide to the expression of uncertainty in measurement (GUM), which evaluates the uncertainty of a measured result through the combination of each source of uncertainty in the measuring process and the approach derived from control chart techniques. The objective of this article is to determine if these two approaches are equivalent, or if in the case of gas chromatography of natural gas, there are differences between them.

KEYWORDS: Measurement uncertainty; GUM approach; Control charts; Natural gas.

---

Introduction

The evaluation of uncertainty associated with an analytic result is an essential part of the measurement process. The uncertainty of a measurement is defined as a parameter associated with the result of a measurement, which characterizes the dispersion of values that can be fundamentally attributed to a measurand [1]. The result of a measurement is considered as the best estimate of the value of measuring accompanied with all the sources of uncertainty that contribute to its propagation [2]. Consequently, the result of a measurement cannot be correctly interpreted without knowledge of the uncertainty of the result [2].

Several concepts have been developed for the evaluation of uncertainty related to the result of a measurement. The approach most used is the one proposed by GUM [3] for the expression of uncertainty in measurements, which combines the diverse sources of uncertainty, by expansion of the Taylor Series. In the beginning of the 1990s, EURACHEM [5] adopted GUM for analytical chemistry.

The control chart as a graphical means of applying the statistical principles of significance to the control of a production process was first proposed by Dr. Walter Shewhart in 1924 [4,5]. Control chart theory recognizes two types of variability. The first type is random variability due to chance causes (also known as common causes). This is due to the wide variety of causes that are consistently present and not readily identifiable, each of which constitutes a very small component of the total variability but none of which contributes any significant amount.

Nevertheless, the sum of the contributions of all of these unidentifiable random causes is measurable and is assumed to be inherent to the process. The elimination or correction of common causes requires a management decision to allocate resources to improve the process and system.

The second type of variability represents a real change in the process. Such a change can be attributed to some identifiable causes that are not an inherent part of the process and which can, at least theoretically, be eliminated. These identifiable causes are referred to as assignable causes or special causes of variation. They may be attributable to the lack of uniformity in material, a broken tool, workmanship or procedures or to the irregular performance of manufacturing or testing equipment.

When we observe that the second type of variability does not exist, the first one – random errors – can be considered as the process uncertainty.

A standard material or control sample is measured repeatedly over time. The presumption of this practice is that the variation experienced on this material will be indicative of the laboratory total expected variation. Incorporation of specific known or potential sources of variation in the testing program is encouraged.

A control chart is prepared and the results are evaluated to identify short-term variation and longer-term variation. These data can then be used to determine an estimate of uncertainty standard deviation.

Methodology

After calibration, over a period of five days, a certified reference material (CRM) was analyzed by gas chromatography. This sample was a natural gas composition. We obtained 20 results, four replicates per day.

The outlier data were tested by Grubbs’ test [6] and the data normality was evaluated using the Anderson-Darling statistic [7].

GUM approach

The GUM approach (also known as the bottom-up approach) consists of the identification and quantification of the relevant sources of uncertainty following the combination of these individual estimates. In the estimation of total uncertainty, it is necessary to deal separately with each source of uncertainty to know its contribution. Each one of the separate contributions of uncertainty is referred to as a component of uncertainty.

When expressed as a standard deviation, a component of uncertainty is known as a standard uncertainty, u(x). If there is a correlation between some of the components, then we must calculate the co-variance. However, it is frequently possible to evaluate the combined effect of the various components.

For a result of measurement x, standard combined uncertainty, uc(x) is an estimate of combined standard deviation equal to the positive square root of the total variance obtained by the combination of all the components of uncertainty evaluated, using the law of propagation of uncertainty.

For many propositions in analytic chemistry [8], an expanded uncertainty, U, may be used. The expanded uncertainty comes from the interval in which the value of the measurand is believed to be with a certain level of confidence. U is obtained by multiplying uc(x), the combined standard uncertainty, by a coverage factor k. The choice of the factor k is based on the required level of confidence.

In gas chromatography, when a quantitative analysis is carried out, the sources of uncertainty identified have to be taken into consideration, both in the calibration and in the sample. Normally, in gas chromatography, due to the difficulty in obtaining certified reference materials, with different concentrations of components, calibration of a single point is utilized. With a view to minimizing the possible systematic errors, the CRM must present the closest possible concentration to the sample being analyzed. For the test considered here, the approach used follows ISO 6974-2 [9]. The mathematical model used to calculate the non-normalized molar fraction is the following:

(1)

where:

i is the non-normalized molar fraction of component i;

xCRM,i is the molar fraction of component i in the CRM, found on the certificate;

CRM,i is the molar fraction of component i in the CRM, analyzed;

i is the molar fraction of component i, analyzed.

The uncertainty of the non-normalized molar fraction of component i , i , from equation (2), in considering the independent quantities among themselves, is:

(2)

where:

is the standard uncertainty of the non-normalized molar fraction of the component i;

uxCRM ,i is the standard uncertainty of the molar fraction of component i in the CRM, found on the certificate;

is the standard uncertainty of the molar fraction of component i in the CRM, analyzed;

is the standard uncertainty of the molar fraction of component i, analyzed.

The mathematical model used to calculate the normalized molar fraction is the following:

(3)

The mathematical model used to calculate the normalized molar fraction is the following:

xi is the normalized molar fraction of component i;

w is the non-normalized molar fraction of each component i.

The uncertainty of the normalized molar fraction of component i , xi, from equation (4) is:

(4)

Control chart approach

Preliminary control charts are then prepared and examined. These charts are evaluated to determine if the process is in a state of statistical control. The usual principles of control charting use short-term variability to estimate the limits within which samples of test results should vary. For control sample programs this short-term variability is equivalent to repeatability precision. It is expected, however, that additional contributions to variation will be present over time and therefore additional variation, equivalent to intermediate precision, will be encountered.

Two types of control charting methods are recommended to develop estimates of uncertainty. These include:

– mean and range or standard deviation charts are used when multiple test results are conducted in each time period;

– individual charts are used when single test results are obtained in each time period.

Either a range chart or a standard deviation chart may be used to estimate the short-term variability when multiple assays are conducted under repeatability conditions per time period.

An estimate from the control chart data can be compared to other estimates of repeatability (within a laboratory, short-term variation) if available.

Sample averages are examined and may provide estimates of variation caused by other factors. Such factors may include environmental effects, operator factors, reagents or instruments. In general, these sources of variation will be defined (including acceptable tolerances) by the test method.

A specified number of independent test results are taken during each time period.

Either a range chart or a standard deviation chart is prepared. This is examined for special cause variation. If the variability appears random then an estimate of repeatability is calculated. This may be done by pooling the sums of squares, using the average standard deviation, or using the average range.

A means chart is used to examine variation among time periods. Limits on this chart enable comparison of variation between time periods using repeatability as the estimate of error.

If the control chart shows a state of statistical control then the uncertainty will be assumed approximately equivalent to the repeatability standard deviation.

In most cases it will be expected that the variability between means will show an out of control condition indicating that there are special causes of variation in addition to repeatability. The between means variation and within means repeatability estimates are then used to calculate an estimate of uncertainty standard deviation.

The following steps are suggested:

1 – Prepare a standard deviation control chart

Calculate the average of the standard deviations (p test periods):

(5)

If the standard deviations in many of the samples were zero, then we recommend replacing the values of 0 with a value calculated as scale division divided by .

2 – Estimate the within sample standard deviation this is an estimate of a single laboratory repeatability standard deviation.

A direct estimate of single laboratory standard deviation, sr , is calculated based on the pooled variances. This is found by: calculating the squares of each standard deviation; summing the squares, ∑ s² , dividing by the number of samples; and taking the square root.

(6)

3 – Estimate the between time or sample variation

Since there is a between sample or between time variation, an estimate of the between time standard deviation is then calculated. First the standard deviation among the sample averages is found.

The stimeis then calculated as:

(7)

where:

= the standard deviation of the averages,

nwithin= number of repeats (4),

swithin = standard deviation within groups and is equivalent, to sr = single laboratory repeatability standard deviation.

Note: If the difference under the radical sign is negative, meaning the estimate of is negative, then this may be interpreted as indicating that the variation associated with time is negligible and the estimate of stime is set to zero.

4 – The uncertainty standard deviation is estimated from a single time and a single repeat

(8)

Note: This value is equivalent to an estimate of intermediate precision based on multiple time periods.

Results and discussion

GUM approach

The GUM approach requires the quantification of all the contributions relevant to the uncertainty.

We have followed method B, from ISO 6974-2, which uses only one calibration point.

We have calibrated the chromatograph using a mixture of standard gas and this CRM was analyzed three times as a sample. This implies the use of linear calibration without an interceptor.

The results of this methodology are found in Table 1, based in the equations (1) to (4).

Table 1. Relative standard uncertainty by the GUM approach

The effect of the normalization algorithm in the GUM approach minimizes the uncertainty of the component with more than 50% in its composition.

With the exception of methane, where the uncertainty of the CRM is the highest source of contribution, as we increase the uncertainty of the CRM, we observe an increase in the uncertainty of the component.

Control chart approach

No value is considered an outlier using Grubbs’ test and the data have normal distribution based on the Anderson-Darling criterion.

The results of this methodology are found in Table 2, based on equations (5) to (8).

Table 2. Relative standard uncertainty by the control charts approach

Finally, we compare both approaches in Table 3. The first approach is separated into two parts: with and without CRM.

Table 3. Comparison of the approaches

When we do not consider the CRM, the results become too similar. This occurs because this source is the largest and the control chart approach can not perceive this variation.

Conclusion

The new approach – control charts – presented compatible results in relation to the GUM results.

This methodology showed itself to be simple to use (without differentials as the GUM), as well as using the same statistical of the process control data.

Acknowledgements

Experimental data were collected thanks to Nilson Menezes Silva and his staff from PETROBRAS/UN/SEAL.

References

[1] International Organization for Standardization, International Vocabulary of Basic and General Terms in Metrology, second ed., International Organization for Standardization, Geneva, 1993.

[2] Analytical Methods Committee, Analyst 120 (1995) 2303–2308.

[3] International Organization for Standardization, Guide for the Expression of Uncertainty in Measurements, International Organization for Standardization, Geneva, 1993.

[4] IS0 7870, Control charts – General Guide and Introduction, 1993.

[5] ISO 8258, Shewhart Control Charts, 1993.

[6] Miller, J. N. & Miller, J. C. Statistics and Chemometrics for Analytical Chemistry, 2000.

[7] ASTM D-6299, Standard Practice for Applying Statistical Quality Assurance Techniques to Evaluate Analytical Measurement System Performance, 2006.

[8] EURACHEM, Quantifying Uncertainty in Analytical Measurement, first ed., EURACHEM, 1995.

[9] ISO 6974: Natural Gas – Determination of Composition with Defined Uncertainty by Gas Chromatography, Part 2, 2001.

---

Performance characteristics of GF-AAS and some metrological aspects of trace element determination

Steluta Duta

Institute for Reference Materials and Measurements, EC-JRC-IRMM, Retieseweg 111, 2440 Geel, Belgium, [e-mail]/[e-mail]

---

ABSTRACT: The performance characteristics of atomic absorption spectrometers used for determination of metal pollutants in water are set-up by legal metrology recommendations. As estimation of results is based on the instrumental comparisons between the signal provided by a known reference solution and the sample under investigation, the instrumental performance characteristics of the equipment are important to provide proper quality results. Some metrological characteristics of the instrument are presented and evaluated. From the measurement uncertainty evaluation of metal pollutant determination in water as well as from the metrological verification/calibration activities of atomic absorption spectrometers, some aspects affecting the quality of results are also identified and discussed.

KEYWORDS: GF-AAS, Trace elements in water, Metrology in Chemistry, Measurement uncertainty

---

1. Introduction

The determination of metal pollutants in water by atomic absorption spectrometry with graphite furnace (GF-AAS) is a common and well established analytical technique for many chemical testing laboratories. There are the standardized procedures [1] as well as other analytical procedures described in the instrument manual from the manufacturers.

The atomic absorption spectrometers with graphite furnace used for metal pollutant determination in water should fulfill some criteria that are set-up by legal metrology recommendations [2]. At the national level, some legal metrological norms [3] and some specific procedures [4] to calibrate the atomic absorption spectrometers are issued and applied on the basis of OIML recommendations.

It is also well known that for accreditation purposes the chemical testing laboratories should fulfill some technical requirements according to ISO/IEC [5]. It is mentioned in this document that ‘equipment shall be calibrated or checked to establish that it meets the laboratory’s specification requirements and complies with the relevant standard specifications’. The measurement uncertainty evaluation and assuring the traceability of measurement results are also required.

In this context, the paper presents some aspects concerning the metrological performance characteristics of atomic absorption spectrometers with a graphite furnace in connection with their consequence on the quality of the analytical results.

2. Performance characteristics of GF-AAS

The traditional metrological assurance of spectrochemical measurement results is mainly concerned with the qualification of instrumental performance characteristics of atomic absorption spectrometers that includes the verification and calibration of the equipment. The proper use of the certified reference materials to assure the traceability of measurement results as well as the evaluation of measurement uncertainty are also important activities.

Accordingly, with the legal metrology recommendations [2, 3], the atomic absorption spectrometers used for metal pollutant determination in water are subject to the compulsory periodical verification of their metrological performance characteristics. Additionally, the authorized metrological laboratories perform the calibration of these instruments at the environmental request of laboratories.

The instruments are tested in a standard configuration and under standard operating conditions to determine the performance characteristics. This is obviously the case for flame atomic absorption spectrometers. For atomic absorption spectrometers with a graphite furnace the experimental conditions should be carefully selected and optimized.

The main instrumental performances of atomic absorption spectrometers with a graphite furnace that are verified are: (i) the working range; (ii) the short term precision; (iii) the limit of detection and (iv) the characteristic concentration (or sensitivity).

As is well know in the atomic absorption technique, the concentration of an unknown sample is evaluated from the calibration data. Calibration means the set of operations that establish, under specified conditions, the relationship, within a specific range, between values indicated by the instrument and the corresponding values assigned to reference standard solutions. The instrument is usually calibrated as far as possible according to the manufacturer’s procedure.

Copper is the recommended chemical element used to evaluate the instrumental performances of atomic absorption spectrometers. A copper standard solution with the nominal concentration (1000±3) mg/L is used as a stock standard solution. The corresponding standard solutions in the working range (2-40) μg/L are prepared by successive dilution steps from working solutions that normally have the concentration in the range (10-100) mg/L.

A spectrometer type SOLAAR 939 with GF 90 furnace and video-auto-sampler SFTV is used. The measurement conditions are wavelengths 324.8 nm; bandpass 0.5 nm; lamp current 80% mA, sample volume 20 μL, D2 background correction. Table 1 shows the optimized temperature programme.

Table 1. Temperature programme for copper determination

An optimal working range is considered as the range where the copper concentration varies proportional with the instrumental signal. The copper standard solutions with the concentration in the range (4-40 μg/L) are used to investigate the working range.

Calibration, evaluated as the regression line of concentration and corresponding absorbances of up to five copper standard solutions is performed. In practice there are two possibilities to perform the calibration, concentration versus height signal or concentration versus area signal. The calibration line and the regression coefficient are shown in Figure 1.

The atomization process is well described either by the height, respectively by the area